Keen to illustrate the spectacular potential of generative AI to his audience at a British Chamber of Commerce in Hungary event in September, Kiss Googled “Laszlo Gaal non-existent car show” and played the video on a big screen.

A highly professional, animated report followed, just 72 seconds long, featuring short interviews with attendees at a swish electric-car exhibition that looked like it had been filmed at a major exhibition center in North America.

“I don’t know this guy,” Kiss said, “I just found this on YouTube,” as the video played, minus the sound (even the most high-tech IT companies encounter glitches, it seems).

“So, this is an e-car show, right?” Kiss continued enthusiastically, only to add: “But it’s not. This show never happened. These people do not exist.” Those watching were agog.

Kiss, a veteran of the Hungarian IT sector and the founding chairman of Attrecto, a software development company with a 15-year track record, reinforced his point.

“It’s not even sophisticated. It’s not a team of marketing guys [that created this]. It’s just one person trying out the Veo3 app the day after it came out last May,” he said.

Indeed, the explosion of fictitious videos this year has profound implications, especially in the political sphere.

Compromised Reality

“Today, if I open my Facebook page, it’s full of AI-generated, politically motivated videos. One is of [Prime Minister] Viktor Orbán riding a zebra, another is of [opposition Tisza Party leader] Péter Magyar cutting the migrant fence. […] So, our basis to even decide about reality, what is real and what is not, is compromised in both image and sound,” Kiss cautioned.

Naturally, he emphasized, the use of AI need not be nefarious. It can, and increasingly is, used to streamline business operations and even bypass expensive “traditional” IT processes.

As an example, Kiss argued that today, someone with no experience whatsoever in writing computer code can nevertheless create software by using a combination of AI tools and the right prompts and instructions.

Although the process is somewhat bewildering for someone unfamiliar with the terms and tools of AI, Kiss indeed produced a management software program and accompanying website without writing any code himself, all within the target time of just one hour.

When asked how this compared to “classical” code writing, he said it might take a team of three or four software developers between four and six weeks of work, costing perhaps EUR 20,000.

This method, dubbed “vibe coding,” a term coined by Slovak-Canadian computer scientist Andrej Karpathy in February this year, has, in just seven months, developed “a whole industry with education, books, workshops, everything,” Kiss said.

However, he stressed that while vibe coding offers great potential, it is still in its infancy and cannot produce a product to commercial standards.

Proof of Concept

“I’m not saying that in 60 minutes we will have a [saleable] product. We will have something that looks like a product, which is either a prototype or a proof of concept […] It’s great [but] it’s not good [enough] to produce production-grade code,” he underlined

Vibe coding is, naturally, just one aspect of a plethora of possible applications emerging that use AI. Sensing this potential, companies and states around the world are pouring money into research to stay ahead of the development curve, none more so than the United States.

By way of illustration, Kiss said that, in 2019, Microsoft invested USD 1 billion into OpenAI, the California-based AI company, creating a “huge story,” only to inject a further USD 10 bln in 2023.

“This was even a bigger story, because at the same time, the European Union fund for developing AI was EUR 6 bln. So, in America, one company invested more in one other company than the complete budget of Europe in AI,” he said.

“Now, OpenAI [and its partners] is committed to build hardware for USD 500 bln, just two or three years later. This is such a big number. I don’t know much money that is. I don’t know how many houses or Ferraris you can build for that.”

Yet even these sums are dwarfed by plans by Anthropic, another U.S. AI startup, backed by Amazon. According to Kiss, this mega-project involves a sprawling data center across some 1,200 acres (500 hectares) housing 30 different buildings, each bigger than a football stadium. All this for one AI computer system, which will need a nuclear power station to meet its electricity needs.

“It is absolutely mind-blowing what is happening in AI,” Kiss told his audience. “It’s hard to understand, even with the facts in front of your eyes.”

How a Hungarian Software Firm Adopts AI to Enhance Productivity

After 10 years working in IT for multinationals, including Siemens, Audi and SAP, in 2010, Gergely Kiss grew tired of his daily commute. Along with two former SAP colleagues, he set up Attrecto in his home city of Győr, 125 km west of Budapest by road.

“In Latin, Attrecto means ‘touch,’ [so-named] because we created the company to work with smartphones, which were brand new at the time. That smartphone world fired our imagination, because, from an engineering perspective, those devices are wonderful, with their [then novel] internet access,” he tells the Budapest Business Journal.

The tiny startup set to work developing specialist apps for clients to enhance their user experience. Before long, those same clients began asking for web development, ultimately pushing Attrecto into what could be termed proto-AI projects, such as chatboxes.

“Around 2015-16, we started AI. At that time, there was no ChatGPT; there was computer vision, algorithms, and then we started to build chatbots. In 2016-17, we built several, mostly for human resources purposes,” Kiss says.

Since the advent of ChatGPT in late 2022, work has increasingly focused on using AI to enhance company productivity.

“Nowadays, our clients often ask: how can I leverage AI in my business process? First, we need to understand that process, then think about how AI can be used. [For example] there is a business process, and you have to create 26 Word documents each day, because […] you are an inspector. Generative AI can create that document; you don’t have to,” Kiss explains.

Attrecto now boasts an impressive list of clients, including Erste Bank, AIG and Mercedes-Benz. It has a team of some 75 full-time and regular contract employees, with revenues of around HUF 1.3 bln expected this year, half of which originates from abroad. Impressive? Kiss isn’t so sure.

“I’m not impressed! The world is really our world, but still, we have this relatively small company. We have been doing this for 15 years, and I always have the feeling that we could have been much better businesspeople and much more creative.”

This article was first published in the Budapest Business Journal print issue of October 17, 2025.